ICES Journal of Marine Science, 53: 691–700. 1996
Size-related mortality of herring (Clupea harengus L.) escaping
through a rigid sorting grid and trawl codend meshes
Petri Suuronen, Jose A. Perez-Comas, Esa Lehtonen,
and Vesa Tschernij
Suuronen, P., Perez-Comas, J. A., Lehtonen, E., and Tschernij, V. 1996. Size-related
mortality of herring (Clupea harengus L.) escaping through a rigid sorting grid and
trawl codend meshes. – ICES Journal of Marine Science, 53: 691–700.
Studies on herring, Clupea harengus (L.), were conducted in the northern Baltic to
assess whether the mortality of herring escaping through a rigid sorting grid (12 mm
bar spacing) placed in front of the codend was different from that of herring escaping
from a 36 mm diamond mesh codend. Escapees were collected into netting cages, and
subsequently transferred into large holding cages (85 m3) where they were held for up
to two weeks to assess mortality. 76–100% of small (<12 cm) and 44–83% of large
(12–17 cm) escapees were dead after 7 days. The 14-day mortalities were 96–100% and
77–100% for small and large escapees, respectively. In the spring experiments, survival
of large herring that escaped through the grid was 7–18% higher than that of codend
escapees and, in the autumn experiments, was 2–7% higher. Survival of small escapees
was not improved by the use of a sorting grid. Observations on escapees suggested that
they suffered considerable loss of scales during capture, and that injured skin areas
were often seriously infected within a few days of escape. Herring caught by seine and
handline, used as controls, were generally in good condition, but they also experienced
mortality. After two weeks their cumulative mortality reached 9% in spring and about
55% in the autumn. On the basis of our results, we argue that a considerable part of the
mortality observed in herring escapees may be attributed to mechanical contacts with
the trawl netting and exhaustion during the capture process, and that damage can
occur in the net long before fish reach the codend.
? 1996 International Council for the Exploration of the Sea
Key words: pelagic trawl, sorting grid, codend mesh, herring, escape mortality,
experimental design.
Received 20 April 1995; accepted 4 July 1995
Petri Suuronen (corresponding author)*, Esa Lehtonen and Vesa Tschernij: Finnish
Game and Fisheries Research Institute, P.O. Box 202, FIN-00151 Helsinki, Finland.
Jose Antonio Perez-Comas: University of Washington, Fisheries Research Institute,
Box 357980, Seattle, WA 98195, USA.
Introduction
Poor trawl selectivity is one of the current concerns in
fisheries management. In order to improve the selective
characteristics of trawl nets, codend modifications are
currently under investigation worldwide. One fundamental aspect of this research is to ensure that fish which
pass out of the trawl net survive, and thus help to build
up the future stock. Treschev et al. (1975) and Efanov
(1981) found that about 90% of Baltic herring (Clupea
harengus L.) survived after escaping from codends of
24–32 mm mesh sizes, although mortality of the smallest
*Present address: University of Washington, Fisheries Research
Institute, Box 357980, Seattle, WA 98195, USA.
1054–3139/96/040691+10 $18.00/0
juveniles was higher. Suuronen et al. (1996), however,
suggested that survival of herring escaping from trawl
codends was 10–40%, and that survival did not depend
on the codend mesh size. They argued that skin injuries
and exhaustion while herring are inside the trawl are
mostly responsible for this high mortality.
The use of rigid grids for sorting undersized fish out
of trawls has been developed recently in Scandinavia
(e.g. Larsen and Isaksen, 1993). According to numerous
underwater observations, the passage of herring through
a rigid sorting grid appears much easier than swimming
through diamond or square meshes (Suuronen, 1991;
Suuronen et al., 1993). The loss of scales of meshselected small haddock (Melanogrammus aeglefinus) was
found to be significantly higher than that of grid-sorted
? 1996 International Council for the Exploration of the Sea
692
P. Suuronen et al.
Trawl
Release
rope
Cover
Cage
Codend
Cage
Sorting
grid
Cover
Codend extension
Guiding panel
Figure 1. General design of a herring midwater trawl with rigid sorting grid and collecting cover and cage. The sorting grid was
mounted at the front upper panel of the codend extension parallel to the netting.
fish (Soldal et al., 1991). Furthermore, a sorting grid
placed in front of the codend may have better selective
performance at higher catch rates because it is not easily
blocked by catch (Suuronen et al., 1993). Thus, given the
differences between both sorting processes, the frequency and degree of injuries in escapees, and their
subsequent mortality, might be expected to be lower for
grid-sorted fish. Thus, sorting grids might be an alternative to conventional mesh selection for fragile and easily
injured fish such as herring.
This paper describes the technique and results from
the survival experiments conducted in spring and
autumn 1993 to assess and compare the mortalities of
herring escaping through rigid sorting grids and codend
meshes.
Materials and methods
Survey area, fishing trials, and sorting devices
Experiments were conducted from 18 May to 10 June
(spring) and from 6 October to 25 November (autumn)
of 1993, in 35–60 m-deep waters in the Archipelago Sea,
northern Baltic Proper, where salinity is approximately
0·6‰. Hauls were performed with a standard herring
midwater trawl (see Suuronen and Millar, 1992). During
the spring, the trawl was towed by the stern trawler
‘‘Harengus’’ (300 hp), and in autumn it was towed by
the ‘‘Mia’’, a traditional side trawler of 850 hp. The
headline height of the trawl during towing ranged from
18–20 m and average towing speed was 3.0 knots (&0.3
knots). In spring, the tows were made in the evening and
at night, and in autumn during the late afternoon and
evening, at depths where most of the herring were
concentrated (usually 10–40 m). The duration of tows
was restricted to 2–10 min to prevent potential injury
induced by the cage after escape (see Suuronen, 1991;
Suuronen et al., 1996). Due to the short towing times,
catches were low (around 30–50 kg). Catches consisted
almost entirely of herring.
We examined the mortality of herring escaping from a
36 mm diamond mesh codend (mesh opening 33–34 mm,
250 meshes in circumference) made of polyamide (PA)
twine (210/30) and through a rigid sorting grid made of
anodised aluminum. The grid consisted of three rectangular parts (50#80 cm each; bar diameter=12 mm,
between-bar distance=12 mm) to ease its bending on the
net drum. It was mounted in the front upper panel of the
codend extension piece parallel to the netting (Fig. 1)
only during those tows aimed to collect herring escaping
through it. The selectivity performance of the 36 mm
diamond mesh codend and the 12 mm grid have been
discussed by Suuronen and Millar (1992) and Suuronen
et al. (1993).
Size-related mortality of herring
(A)
693
(B)
Rope
Cover
Cage
Cage
Codend
(C)
Cage
Holding
cage
(D)
(E)
(F)
Holding
cage
Cone
Figure 2. Schematic representation of the experimental procedures used during towing (A), release (B) and transport of collection
cage (C), and during transferring of fish from collection cage to large holding cage (E), adjusting the depth of holding cage (D),
and collecting of dead fish with a netting cone (F).
Survival experiments
Collection of escapees and caging techniques
A hooped netting cage (diameter 2 m, length 7.5 m,
volume approximately 20 m3) was attached to a cover
fitted over the codend or the sorting grid (Figs 1, 2A).
Fish that escaped from the codend or sorting grid were
guided by the cover aft into the cage. The cover and the
cage were made of unimpregnated knotless (14 mm
stretched mesh length) nylon netting except for the
rearmost panel of the cage which was made of 6 mm
stretched netting. The meshes were hung in a square
orientation. During tows, the cover was held away from
the codend meshes by means of three hoops (diameter
2 m) giving the cover a cylindrical form. After 2–10 min
of towing, the cage was remotely released from the trawl
and closed (Fig. 2B), and slowly transported at Y knot
speed in about 5–10 m depth to the large holding cages
(Fig. 2C). The towing direction and release position
were chosen to minimise transport time to the cage site.
Transport time varied from 5–35 min.
The 85 m3 underwater holding cages were also made
of unimpregnated 14 mm (stretched length) knotless
nylon meshes hung in a square. Holding cages had a
diameter of 4 m and a total height of 10 m (Fig. 2D).
Figure 2e illustrates the technique used when transferring the fish from the towing cage into the large holding
cage. The mouths of the two cages were connected
together by zippers and the whole system was carefully
pulled down so that the fish could swim freely from the
upper towing cage to the lower holding cage. Finally, the
system was lifted, and the upper cage was detached from
the large cage which was then closed and lowered to the
chosen depth.
Usually, escapees from two successive hauls were
transferred into a holding cage to get enough fish in all
size-classes and to reduce inter-haul variation. A total of
seven holding cages (three for codend escapees, four for
grid escapees) were deployed in spring, while eleven
cages (eight for codend escapees, three for grid escapees)
were used in autumn. The duration of the caging period
was usually about two weeks (Table 1, T).
Holding cages were anchored in sheltered areas to
reduce the risk of storm damage. However, in autumn,
occasional strong sea currents tipped some cages into
almost-horizontal orientation. To minimise possible
694
P. Suuronen et al.
Table 1. Starting day of experiment, treatment, tow duration, cage depth (uppermost part), caging period (T), observations (n),
initial number of small (NS) and large (NL) fish kept in holding cage and subset code. (C=coded, G=grid, H=hook and line,
S=seine, Sp=spring, A=autumn, Sm=small herring, Lg=large herring).
Date
Treatment
Tow time
(min)
Depth
(m)
T
(day)
n
1
2
3
4
5
6
7
8
18/5/93
19/5/93
25/5/93
27/5/93
1/6/93
7/6/93
9/6/93
10/6/93
Grid
Codend
Grid
Grid
Codend
Hook
Grid
Codend
11.5
10
10.5
10
10
—
11
11.5
7
7
7–8
7–8
7–9
8–13
8–10
8–10
6.8
12
15.3
13.9
14.9
24.1
14.4
13.5
3
5
4
4
5
12
8
6
9
10
11
12
13
14
15
16
17
18
19
20
21
22
10/6/93
10/7/93
14/10/93
15/10/93
21/10/93
22/10/93
28/10/93
29/10/93
3/11/93
3/11/93
11/11/93
18/11/93
24/11/93
25/11/93
Seine
Seine
Codend
Grid
Codend
Codend
Codend
Codend
Codend
Codend
Hook
Grid
Grid
Codend
—
—
10
10
5
5
5
5
5
5
—
5
3.5
2
5
0
5
5
0
2
2
0–2
0
2
2
2
2
2
14
13.1
13.8
14
13
12.1
13.9
16.9
8.9
12
19.9
12.9
13.9
12.9
5
5
8
7
7
6
9
10
6
6
10
7
8
8
Cage
differences in environmental conditions during the
observation period, cages were anchored close to each
other. In the spring, cage depth was chosen so that the
mid-section of the cage was always in waters of 10)C
(&1)C). Usually, the uppermost part of the cage was at
7–10 m depth (Table 1). In the autumn, temperature of
the upper water layer (0–30 m) varied from 3–5)C.
Nevertheless, cage depth was varied so that the fish of
three cages were allowed access to the surface (Table 1).
Predation of fish by sea birds was prevented by a framed
netting roof.
Dead and moribund fish were removed and counted,
on average, every second day after capture (Table 1, n).
Each time, after lifting the cage near to the surface, these
fish were collected with a netting cone mounted in
the cage (Fig. 2F). The total length of each fish was
measured to the nearest half cm, and a visual inspection
for superficial injuries was performed. On completion of
each survival experiment, the whole cage was lifted to
the surface, all dead and live fish were counted and
measured, and their injuries inspected. The pooled
length distributions of codend and grid herring escapees
from the spring and autumn experiments are shown in
Figs 3a, 3c, respectively. With the exception of the
spring grid escapees, the distributions were markedly
bimodal.
Gear performance during towing and fish behaviour
during transport and caging was occasionally observed
by an underwater camera.
NS
NL
Subset
76
359
58
287
60
—
49
85
89
341
78
808
200
259
89
287
GSmSp
GLgSp
CSmSp
CLgSp
GSmSp
GLgSp
GSmSp
GLgSp
CSmSp
CLgSp
HLgSp
GSmSp
GLgSp
CSmSp
CLgSp
758
560
196
18 896
44
136
158
468
68
43
—
40
5
48
—
—
133
286
3
18
86
48
10
231
161
307
34
4
SSmA
SSmA
CSmA
CLgA
GSmA
GLgA
CSmA
CLgA
CSmA
CLgA
CSmA
CLgA
CSmA
CLgA
CSmA
CLgA
CSmA
CLgA
HLgA
GSmA
GLgA
GSmA
GLgA
CSmA
CLgA
Fish caught by handline and seine
As controls, cumulative mortalities of non-trawled
herring held in similar caging conditions were also
estimated. On two occasions (7 June and 11 November),
herring were caught by handline fitted with barbless
hooks (non-baited) at depths of 5–10 m at dusk, and
caged in large holding cages for 20–24 d (Table 1). Fish
which were hooked cleanly on the lips and showed no
apparent signs of physical damage were immediately
released into the holding cage that had been secured
along-side the boat. These fish did not experience any
transport, neither were they touched by hand. Only
herring larger than 12 cm were caught by hook and line
(Fig. 3b, 3d). In autumn, herring were also caught by a
small purse seine made of small mesh knotless nylon
netting. These were primarily small herring (Fig. 3d)
which were transferred to two large holding cages and
held for 13–14 d (Table 1).
Data analysis
Given the bimodality shown by herring length frequencies (Fig. 3) and the knowledge of possible differences between mortality rates for small and large herring
(Suuronen et al., 1996), we calculated cumulative mortalities for two length groups. One, subsequently termed
‘‘small’’, consisted of all herring smaller than 12 cm,
while the ‘‘large’’ group included fish of 12–17 cm. Fish
larger than 17 cm were excluded from the analysis.
Size-related mortality of herring
200
695
200
n
(a)
(b)
150
150
100
100
50
50
0
5
7
9
11
13
15
17
19
21
600
7
9
11
13
15
17
(c)
19
21
(d)
500
500
400
400
300
300
200
200
100
100
0
5
600
Mode = 6365
n
0
5
7
9
11 13 15
Length (cm)
17
19
21
0
5
7
9
11 13 15
Length (cm)
17
19
21
Figure 3. Length frequency distributions for herring mortality spring (a and b) and autumn (c and d) data. For (a) and (c):
=codend. For (b) and (d): =seine; =hook.
Cumulative mortalities at each observation time t for
length group g and experimental cage i were calculated
as Mg,i(t)=dg,i(t)/Ng,i, where Ng,i is the total number of
fish of size group g initially placed in holding cage i
(Table 1, NS and NL), and dg,i(t) is the cumulative
number of dead fish at observation time t. Observation
time t was measured as decimal fractions of days,
counted from the start of the caging period (t=0).
For our analysis, cumulative mortalities were classified into 11 subsets (Table 1) according to sorting
devices or catching technique (C=codend, G=grid,
H=hook and line, and S=seine), season (Sp=spring and
A=autumn), and fish size group (Sm=small and
Lg=large). For example, CSmA is the subset of all
cumulative mortalities estimated at each observation
time for each of the eight cages deployed in autumn that
contained small herring that had escaped through
codend meshes.
Observations and collection of dead fish from holding
cages had been planned to occur every 24 h, counted
from the time at which cages were deployed. Unfortunately, the long distance between cage locations and the
harbour and weather conditions did not allow us to keep
such a strict schedule. Thus, observation time t varied
considerably among experimental cages, creating a
design with many empty cells. To avoid this problem
and still be able to assess cumulative mortality after
one and two weeks of caging, we decided to treat
=grid;
observation time t as a continuous explanatory variable
and model cumulative mortality M as a logistic function:
á
M(t)=
1+exp(â"ã#t)
where á, â, and ã are parameters such that á¡1, â>0,
and ã>0. Model fits for each of the 11 subsets were
obtained by non-linear least squares using available
software.
We produced 1000 bootstrap replicates for each subset by resampling from the ordered residuals of the
original fits. Bootstrap replicates were also fitted to the
logistic model, and predicted cumulative mortalities at
time t=7 and 14 days (M
| (7) and M
| (14)) were calculated.
The 2.5 and 97.5 percentiles of the sorted distributions
of M
| (7) and M
| (14) were used as 95% confidence interval
limits. Standard errors for the parameter estimates were
also based on the bootstrap samples.
Results
Fish behaviour and condition during experiments
Observations of the escapees during transportation
(Fig. 2C) to the cage site revealed that most of them
swam calmly within the cages. Some fish showed
occasional bursts of swimming. However, fish were not
seen to have vigorous contacts with the cage. When
herring were transferred from the towing cage to the
696
P. Suuronen et al.
Table 2. Parameter estimates and standard errors (between parentheses) for logistic mortality curves. Average percent cumulative
mortality (M
| ) after 7 and 14 days of cage confinement and corresponding 95% confidence intervals. Average cumulative
mortalities, standard errors, and confidence intervals are based on 1000 bootstrap samples.
Subset
á
â
ã
M
| (7)
M
| (14)
CSmSp
0.9974
(0.0075)
0.8387
(0.0202)
0.9856
(0.0021)
1.0000
(0.0031)
0.8888
(0.0324)
0.7170
(0.0489)
0.9832
(0.0041)
1.0000
(0.0120)
0.1075
(0.0022)
0.7068
(0.0302)
0.4765
(0.1290)
1.5255
(0.3133)
8.8286
(1.5764)
2.5515
(0.2346)
3.2092
(0.4616)
8.8990
(3.7468)
4.5937
(0.6552)
2.3323
(1.0855)
2.5820
(0.2926)
4.2699
(0.2780)
4.9366
(0.4976)
6.5988
(8.3586)
0.3777
(0.0518)
1.2669
(0.2216)
1.6221
(0.1527)
0.6594
(0.0756)
2.4983
(1.1458)
0.8572
(0.1449)
1.1789
(0.3661)
0.3979
(0.0434)
0.4217
(0.0283)
0.4521
(0.0501)
1.0466
(1.2384)
76.33
70.88–82.31
43.72
35.27–51.87
99.81
99.29–99.99
82.69
77.09–88.65
94.94
83.34–99.99
54.66
43.43–65.81
99.30
98.03–99.94
57.69
52.96–62.69
2.36
1.97–2.70
11.70
7.84–14.95
31.84
21.95–43.60
97.76
95.71–99.05
89.94
85.62–92.86
99.84
99.32–100.00
99.57
98.63–99.91
96.40
88.90–100.00
77.43
68.47–86.10
99.74
98.91–100.00
95.33
91.79–97.67
8.93
8.59–9.27
56.13
51.64–59.87
55.31
44.18–65.44
CLgSp
CSmA
CLgA
GSmSp
GLgSp
GSmA
GLgA
HLgSp
HLgA
SSmA
large holding cages (Fig. 2E), their behaviour was calm
and they did not attempt to escape. Occasionally, however, some fish were seen to be trapped by the netting
folds of the upper cage, but later they swam into the
lower cage. Some skin abrasion may have occurred on
these occasions. No dead herring were observed during
transfer.
While in the large holding cages, most trawl-escapees
kept swimming within the middle parts of the cages,
usually forming loose shoals. Some fish swam lethargically away from the main school. Generally, the activity
of fish appeared low, and no aggressive behaviour was
observed. No observations were conducted at night and
during the conditions of strong currents. In spring, after
the third or fourth day of captivity, many live fish were
observed with parts of their skin covered by white
mucus. In autumn, severe skin infections were also
observed, usually a few days later than in spring.
Underwater observations demonstrated that during
emptying (Fig. 2F), the netting cone mounted inside the
cages collected all the dead and moribund fish lying on
the bottom of the cage without causing any observable
harm to those alive. Visual examination of the fish
removed at the periodic visits to the cages showed
various degrees of external injury and decomposition.
Some moribund and recently dead fish had much of
their skin infected and covered by mucus, and some
others had part of their skin swollen or peeled off
revealing the muscle. Skin damage generally increased
towards the caudal peduncle, and in many cases fish had
lost the caudal fin. Often 20–40% of the skin of these fish
was abraded. Usually, larger fish had substantially more
visible injuries than the smaller ones. Living fish
removed at the end of the caging period often displayed
small local skin infections, and in some cases had red
sores on their flanks.
Contrary to the condition displayed by trawl-caught
herring, fish caught by handline were usually in good
condition during the three weeks of observation. The
skin of those fish still alive at the end of the caging
period did not display white mucus or injuries; only
minor local infections, mainly on the snout, were occasionally observed. The dead fish displayed various levels
of decomposition. Small herring caught in autumn by
seine were also in good condition during the two weeks
of confinement, and only a few live fish had local skin
infections.
Mortality of cage-kept herring
The scatter plots and fitted curves in Figure 4 (see Table
2 for parameter estimates) show distinct patterns for the
cumulative mortalities of trawl-escapees. While mortalities were generally low on the first caging day, they
increased dramatically after a few days, approaching the
asymptote value, á, which represents approximate maximum cumulative mortality (Table 2). This value was 1 or
very close to 1 for all the small trawl escapees (with the
only exception of GSmSp, á=0.89) and for the large
escapees of the autumn experiments. Asymptotes for the
Size-related mortality of herring
697
1
0.8
M
0.6
Codend Autumn
Codend Spring
0.4
0.2
0
1
0.8
M
0.6
Grid Autumn
0.4
Grid Spring
0.2
0
1
Autumn
0.8
Hook
M
0.6
Seine
0.4
Hook Spring
0.2
0
7
14
t (days)
21
0
7
14
t (days)
21
Figure 4. Final fits for cumulative mortalities of small (solid circles) and large herring (open circles).
large escapees of the spring experiments were lower than
1 (á=0.84 and á=0.72 for CLgSP and GLgSP, respectively). Moreover, with the exception of the spring
codend experiments, small escapees died faster than
large escapees (Fig. 4; compare the steeper slopes of the
curves for small escapees to those for large escapees).
Curve asymptotes for the large fish were approached at
t¢7 days while those for the small herring were reached
in four to seven days (Fig. 4).
Cumulative mortalities for herring caught by seine
and hook showed a different pattern. On the first caging
days they were as low or lower than those for trawlescapees, remaining low on following days. Asymptotes
á were considerably lower than those for trawl-escapees,
especially for the herring caught by hook in the spring
experiments (Fig. 4, Table 2, HLgSp, HLgA, and
SSmA).
We estimated cumulative mortalities after one and
two weeks of caging (M
| (7) and M
| (14)) from the 1000
bootstrap replicates of each subset to provide a good
assessment of the variation of mortality estimates. Most
of the bootstrap distributions for M
| (7) and M
| (14) were
skewed. Those for M
| (7) were either slightly positive or
negatively skewed with only two symmetric distributions
(those for HLgSp and CLgA). Those for M
| (14) were
negatively skewed with the exception of HLgSp. Table 2
698
P. Suuronen et al.
Spring
1
M
0.8
0.6
0.4
0.2
0
CSmSp
GSmSp
CLgSp
GLgSp
HLgSp
Autumn
1
M
0.8
0.6
0.4
0.2
0
CSmA
GSmA
SSmA
CLgA
GLgA
HLgA
Figure 5. Average predicted mortalities at times t=7 and 14 days, and corresponding 95% confidence limits for bootstrap samples.
: M(T=7); : M(T=14).
and Figure 5 show the average and 95% confidence
limits for these distributions. The average mortalities
M
| (7) and M
| (14) were higher for the autumn experiments
(Fig. 5). Those for herring caught by seine or by
handline were noticeably lower than those for codend or
grid escapees. Moreover, average mortalities were lower
for the large escapees, for both the spring and autumn
experiments. Finally, M
| (14) averages were smaller for
grid escapees than for codend escapees, and this pattern
was consistent across fish size and season. M
| (7) averages, on the other hand, showed an inconsistent pattern
(Fig. 5).
Discussion
The results of the present study suggest that (a) most of
the small (<12 cm) herring escapees will die due to the
trawl capture process, (b) large (12–17 cm) escapees have
a better chance of survival, and (c) survival of escapees is
not markedly improved by the rigid sorting grid
mounted in front of the codend.
Based upon our direct and underwater observations
of herring escapees which showed considerable loss of
scales and skin infections, characteristics that were
almost totally absent in seine or handline caught fish, we
argue that these injuries are mainly trawl-related, and
that they greatly contributed to the observed mortality
of escapees. Numerous underwater observations (e.g.
Suuronen and Millar, 1992; Suuronen et al., 1993) have
shown frequent and vigorous contacts of herring with
the net, particularly along the top panels of the funnelling rear part of the trawl, during the catching process,
resulting in heavy skin abrasion and scale loss. Skin
damage in turn may account for mortality by facilitating
invasion of infectious agents and making the fish more
susceptible to disease (e.g. Efanov, 1981; Lockwood
et al., 1983; Hay et al., 1986; Main and Sangster, 1990;
Sangster and Lehmann, 1994). The muscular exhaustion
caused by the herding of the trawl may also contribute
to mortality (Beamish, 1966; Wardle, 1981; Suuronen
et al., 1996), particular within the smaller size group.
Moreover, damage to skin and severe exhaustion may
both cause disturbance of the osmotic balance which
in due time may cause death (e.g. Roald, 1980; Borisov
and Efanov, 1981; Rosseland et al., 1982; Milligan and
Wood, 1986). However, given the low salinities of the
Baltic Sea, osmotic stress may have had a smaller effect
on mortality than if the experiments had been conducted
in the North Sea (Wardle, 1981).
Small herring escapees died substantially faster than
the larger ones, although one would expect lower mortality in small fish because it should be easier for them to
Size-related mortality of herring
pass through meshes or sorting grids. Several other
studies have also indicated higher mortalities for smaller
escapees (e.g. Borisov and Efanov, 1981; Soldal et al.,
1993; Sangster and Lehmann, 1994; Suuronen et al.,
1996). Underwater observations have clearly shown
that, during trawling, small fish, including small herring,
are usually forced to swim beyond the limits of their
stamina (e.g. Main and Sangster, 1981; Suuronen and
Millar, 1992; Suuronen et al., 1996). Furthermore, disturbance of the osmotic balance may be more acute
among small fish because of their higher surface/volume
ratios.
Light conditions during trawling may have influenced
our mortality results since most of the hauls were
conducted during the evening or at night. It is fairly well
documented that the behaviour of fish in relation to
moving netting is primarily determined by their reaction
to visual stimuli. As visibility range decreases during
darkness, herding becomes reduced and, below a threshold level, fish are unable to react in an ordered pattern
(e.g. Blaxter and Parrish, 1966; Blaxter and Batty, 1985;
Glass and Wardle, 1989; Walsh and Hickey, 1993).
Thus, one may expect that herring swimming in the net
during darkness will suffer more abrading contacts with
the netting and find it increasingly difficult to escape
through the meshes or the grid. For example, Suuronen
et al. (1995) observed that the highest mortality of
5–10 cm vendace (Coregonus albula) escaping from a
24 mm square mesh codend was associated with hauls
conducted during darkness.
Herring escapees died faster in autumn than in spring.
This difference suggests some seasonal effect on escape
survival. However, herring caught by handline also
showed a similar mortality pattern. Reasons for this
seasonal difference can only be guessed. For example,
the possible adverse effects of cage confinement could
have been amplified by less than optimal autumn
weather conditions. It is also possible that the substantially longer nights in autumn (at latitude 60)N,
the average night span is 15 h in the beginning of
November, but only 5.5 h in late May) increased
the probability of unintentional encounters between fish
and cage, causing increased skin abrasion and induced
mortality. An additional factor, particularly in the case
of small escapees, could have been the different body size
(see Fig. 3) and physiological condition of fish. On the
basis of our results, it is not clear how much of the
increased mortality in autumn trials can be attributed to
trawl-related factors.
The advantage of implementing sorting grids over
more traditional codend mesh size regulations is not
that obvious from our study. Although previous
underwater observations showed that the passage of
herring through rigid sorting grids was easier than if
they swim through codend meshes (Suuronen, 1991;
Suuronen et al., 1993), the present experiments suggest
699
that the survival of herring escapees will not be greatly
improved by the use of a rigid sorting grid mounted in
the front part of the codend. Only large herring may
benefit somewhat from grid sorting; the average predicted mortality was 13% lower for grid escapees in
spring, and 4% lower in autumn. Among small herring,
differences were minimal. It is questionable whether
such small reductions in the mortality of herring may
have a real impact on the future of the fishery. However, there might still be good opportunities to
improve the design and placement of grids, and thus
attain greater survival of fish escaping through it. For
instance, if the grid was placed before the rear funnelling part of the trawl, where most of the trawl-induced
damage is likely to occur, the survival of escapees
might improve substantially.
Acknowledgements
We owe special thanks to Aud V. Soldal, Ahto Järvik,
and Olle Hagström for several useful suggestions that
led to the improvement of our experiments and to
Daniel Erickson, Don Gunderson, Bill Karp, Taivo
Laevastu, Russell Millar, John Wallace, and Vidar
Wespestad for critical comments on previous manuscript versions. Thanks also to skippers Paavo Tarna
and Seppo Peussa for their help, and to Ari Orrensalo
and Raimo Riikonen for the technical assistance during
the experiments. Finally, we would like to extend our
sincere thanks to the Nordic Council and the Finnish
State Educational Institute of Fisheries for providing
funding and other resources for this study.
References
Beamish, F. W. H. 1966. Muscular fatigue and mortality in
haddock, Melanogrammus aeglefinus, caught by otter trawl.
Journal of the Fisheries Research Board of Canada, 23:
1507–1521.
Blaxter, J. H. S. and Parrish, B. B. 1966. The reaction of marine
fish to moving netting and other devices in tanks. Marine
Research, 1. 15 pp.
Blaxter, J. H. S. and Batty, R. S. 1985. Herring behaviour in the
dark: responses to stationary and continuously vibrating
obstacles. Journal of the Marine Biological Association UK,
65: 1031–1049.
Borisov, V. M. and Efanov, S. F. 1981. To the problem of
natural and traumatical mortality of Baltic herring. ICES
C.M. 1981/J:13. 10 pp.
Efanov, S. F. 1981. Herring of the Gulf of Riga: the problem of
escapement and mechanical impact of the trawl. ICES C.M.
1981/J:7. 16 pp.
Glass, C. W. and Wardle, C. S. 1989. Comparison of the
reactions of fish to a trawl gear, at high and low light
intensities. Fisheries Research, 7: 249–266.
Hay, D. E., Cooke, K. D., and Gissing, C. V. 1986. Experimental studies of Pacific herring gillnets. Fisheries Research, 4:
191–211.
700
P. Suuronen et al.
Larsen, R. and Isaksen, B. 1993. Size selectivity of rigid sorting
grids in bottom trawls for Atlantic cod (Gadus morhua) and
haddock (Melanogrammus aeglefinus). ICES Marine Science
Symposia, 196: 178–182.
Lockwood, S. J., Pawson, M. G., and Eaton, D. R. 1983. The
effects of crowding on mackerel (Scomber scombrus
L.)—physical condition and mortality. Fisheries Research, 2:
129–147.
Main, J. and Sangster, G. I. 1981. A study of the fish capture
process in a bottom trawl by direct observation from a towed
underwater vehicle. Scottish Fisheries Research Reports, 23.
24 pp.
Main, J. and Sangster, G. I. 1990. An assessment of the scale
damage to and survival rates of young gadoid fish escaping
from the cod-end of a demersal trawl. Scottish Fisheries
Research Reports, 46. 28 pp.
Milligan, C. L. and Wood, C. M. 1986. Intracellular and
extracellular acid-base status and H + exchange with the
environment after exhaustive exercise in the rainbow trout.
Journal of Experimental Biology, 123: 93–121.
Roald, S. O. 1980. Net marks on Atlantic salmon (Salmo salar
L.) in Norwegian coastal areas. Preliminary report on gross
histological, serological and bacteriological signs. ICES C.M.
1980/M:34, 6 pp.
Rosseland, B. O., Trygve, B. L., and Hansen, L. P. 1982.
Physiological effects and survival of Carlin-tagged and
descaled Atlantic salmon, Salmo salar L., in different water
salinities. ICES C.M. 1982/M:30, 23 pp.
Sangster, G. I. and Lehmann, K. 1994. Commercial fishing
experiments to assess the scale damage and survival of
haddock and whiting after escape from four sizes of diamond
mesh codends. ICES C.M. 1994/B:38. 24 pp.
Soldal, A. V., Isaksen, B., Marteinsson, J. E., and Engås, A.
1991. Scale damage and suvival of cod and haddock escaping
from a demersal trawl. ICES C.M. 1991/B:44, 7 pp.
Soldal, A. V., Isaksen, B., Marteinsson, J. E., and Engås, A.
1993. Survival of gadoids that escape from a demersal trawl.
ICES Marine Science Symposia, 196: 122–127.
Suuronen, P. 1991. The effects of a rigid grating on the selection
and survival of Baltic herring – preliminary results. ICES
Fish Capture Committee C.M. 1991/B:17, 22 pp.
Suuronen, P. and Millar, R. B. 1992. Size-selectivity of
diamond and square mesh codends in pelagic herring trawls:
only small herring will notice the difference. Canadian
Journal of Fisheries and Aquatic Sciences, 49: 2104–2117.
Suuronen, P., Lehtonen, E., and Tschernij, V. 1993. Possibilities to increase the size selectivity of a herring trawl by
using a rigid sorting grid. Paper presented at the Symposium
on Gear Selectivity/Technical Interactions in Mixed Species
Fisheries hosted by NAFO, Dartmouth, Nova Scotia,
Canada, 13–15 September 1993. NAFO SCR Doc. 93/119,
Serial no. N2313, 12 pp.
Suuronen, P., Erickson, D. L., and Orrensalo, A. (1996).
Mortality of herring escaping from pelagic trawl codends.
Fisheries Research, 25: 305–321.
Suuronen, P., Turunen, T., Kiviniemi, M., and Karjalainen, J.
(1995). Survival of vendace (Coreogonus albula) escaping
from a trawl codend. Canadian Journal of Fisheries and
Aquatic Sciences, 52: 2527–2533.
Treschev, A. I., Efanov, S. F., Shevtsov, S. E., and Klavsons,
U. A. 1975. Die Verletzung und die überlebenschancen des
Ostseeherings nach dem Durchdringen durch die Maschen
des Schleppnetzsteerts. Fischerei Forschung, 13: 55–59.
Walsh, S. J. and Hickey, W. M. 1993. Behavioural reactions of
demersal fish to bottom trawls at various light conditions.
ICES Marine Science Symposia, 196: 68–76.
Wardle, C. S. 1981. Physiological stress in captive fish. In
Aquarium Systems, pp. 403–414. Ed. by A. D. Hawkins,
Academic Press, London.